This invention relates to telecommunication systems, and more specifically, to inverse multiplexing data streams over multiple links.
In telecommunication networks or systems, data or a data stream is transported from one location in the network to another location in the network at various data rates. Thus, the situation may arise, at some point in the network, where the transport data rate for an incoming data stream exceeds the capacity of a single link over which the data stream needs to be transported. Known solutions to this problem teach that the data stream can be distributed or split into separate streams and the separate streams sent over multiple links or lines of lower capacity; the aggregate capacity of the lower capacity links is sufficient to carry the data stream This approach to splitting data or transporting the data stream over several links is known as “inverse multiplexing”.
Even when a high capacity link is available in the network that can handle the entire incoming data stream, the data stream may not make full use of the capacity of the single link. Thus, current standards teach that it may be preferable to inverse multiplex the data stream onto a number of lower capacity links, and thereby fully utilize the capacity of the links in the network.
In a typical network, there are various bandwidths described in terms of data stream rates or bit rates. For example, a DS1 or T1 bit stream is transmitted at a line rate of 1.544 Mbps. The terms “DS1” and “T1” are used interchangeably herein. T1 is a full-duplex system: transmitted signals are transported on one wire pair, and received signals are transported on a separate wire pair. In each direction, the 1.544 Mbps data streams are organized according to a predetermined protocol. An alternative data rate is E1. E1 bit streams are transmitted at a line rate of 2.048 Mbps.
In order to transport data from one location to another, the data is packaged according to a predetermined protocol. One protocol is Asynchronous Transfer Mode (ATM). In accordance with ATM standards, the data is packaged in cells that are called ATM cells. Each ATM cell is 53 bytes in length, wherein each byte is an octet that is made up of eight bits. Each ATM cell includes a payload or information that is 48 octets in length and a header that is 5 octets in length. The header includes information about the payload type (PT) as well as other information. There are various forms of payload, including idle payload. Idle ATM cells may be present in an ATM data stream and may be inserted or deleted by the equipment processing the ATM data stream. ATM equipment that communicates at rates exceeding the capacity of a T1 line can communicated over multiple T1 lines, which have an aggregate capacity comparable to the rate of the ATM equipment, using an inverse multiplexing scheme.
In the inverse multiplexing scheme, the ATM data stream is divided over several low capacity lines, such as the T1 lines. For example, ATM Forum specification “Inverse Multiplexing for ATM (IMA) Specification Version 1.0”, AF-PHY-0086.000 (July 1997) defines one approach to inverse multiplexing ATM cell streams on multiple T1 lines, which is incorporated herein by reference.
Depending on the transmission rate or bandwidth demand of the ATM data stream, the ATM data stream will have to be divided over several lower capacity lines. For example, if the data stream is received at a rate that is four times an optimal data rate of the lower capacity lines, then the incoming ATM data stream will have to be inverse multiplexed onto or carried by at least four lines.
Known methods of inverse multiplexing teach that all of the low capacity links, among which the ATM data stream is inverse multiplexed, have to be trained at an optimal rate and synchronized so that each line is transmitting from the transmitter end to the receiver end at the same rate. Once the optimal rate is determined and selected, which is based on the number of links needed and the rate or bandwidth of each link, then all of the links operate at that optimal rate until one of the links fails and the optimal rate has to be recalculated.
The disadvantage of current solutions, and thus, the problem with the known methods for determining the optimal rate is that the links are considered and trained as needed. This leads to unnecessary time delays for restoring traffic flow when a link failure occurs because the characteristics of other links have to be determined and a new optimal rate established. Thus, at a data rate of 8 Mbps for an ATM cell stream, every minute that there is a delay in restoring traffic flow 480 Mb of traffic stack up or are lost.
Therefore, what is needed is a system and method for determining as many links characteristics as possible in order to calculate a new optimal rate, and hence, eliminate the down time caused by a failure in a link.